Effect of soils with varying degree of weathering and pH values on phosphorus sorption

Catena 139 (2016) 214–219

Contents lists available at ScienceDirect

Catena journal homepage: www.elsevier.com/locate/catena

Effect of soils with varying degree of weathering and pH values on phosphorus sorption V. Antoniadis ⁎, R. Koliniati, E. Efstratiou, E. Golia, S. Petropoulos Department of Agriculture Crop Production and Rural Environment, University of Thessaly, Fytokou Street, GR‐384 46, Volos, Greece

a r t i c l e

i n f o

Article history: Received 9 March 2015 Received in revised form 23 December 2015 Accepted 8 January 2016 Available online 14 January 2016 Keywords: Crystalline oxides Amorphous oxides Sorption maximum Phosphorus saturation index

a b s t r a c t Various soil properties are known to influence P retention, but it is not clear which of them are predominant when soils of different degree of weathering are compared, and which when newly developed soils differ mainly in activity (pH). We chose 23 typical Mediterranean low organic matter-content soils: 13 of them differed in weathering (4 Alfisols, 4 Entisols, and 5 Inceptisols), and 10 were newly developed Entisols, of which 5 were acidic and 5 alkaline. We conducted batch P sorption tests at C0 = 0–100 mg L− 1, measured important soil physico-chemical properties, and correlated them with sorption indices. Alfisols were significantly higher in total “free” Al and Fe, as well as in well-crystalline oxides, and this led to higher P sorption by Alfisols. Correlation analyses in the activity-divided soils revealed that amorphous oxides, although significantly higher in the acidic soils compared to the calcaric, did not have any influence in P sorption, neither did any other oxides species, while CaCO3, along with pH, were the important factors to enhanced P retention. Contrary to that, in the taxonomy-divided soils, neither pH nor CaCO3 played any significant role in P sorption, which was rather influenced by oxides (mainly the amorphous, but to a lesser degree by the crystalline species as well). We conclude that oxides are the key soil property influencing P sorption among soils of different weathering (even if these soils also differ in pH and CaCO3), while within the same taxonomic order, CaCO3 and pH becomes the important factor. © 2016 Elsevier B.V. All rights reserved.

1. Introduction Phosphorus, an important macro-nutrient, is usually the single most restricting element for plant production. The reason is that its readily phyto-available soil pool, the water soluble, is only a minor fraction of the total soil P, which, during high-demand seasons may be replenished many times per day, and thus needs to be recharged by other labile P pools (Nesme et al., 2014). Thus labile P species influence the levels of soil P availability, and are known to be related to certain soil properties, most important of which are pH, CaCO3, organic matter, and Al, Fe, and Mn oxides (Al-Rohily et al., 2013). There are two types of studies typically employed for assessing P availability: sorption tests and the use of extractants. From such studies it has been reported that P tends to be strongly retained by soils with slightly alkaline/calcaric, as well as with acidic pH, each for different reasons. In calcaric soils, P is bound by CaCO3, while in acidic pH it is soil oxides that are expected to play a key role. Carbonate surfaces have a two-fold effect: They provide abundant Ca ions that cause P precipitation as insoluble Ca-P species, and directly retain P

⁎ Corresponding author. E-mail address: [email protected] (V. Antoniadis).

http://dx.doi.org/10.1016/j.catena.2016.01.008 0341-8162/© 2016 Elsevier B.V. All rights reserved.

− electrostatically through the exchange of HCO− 3 for H2PO4 . This becomes more effective as specific surface increases, i.e., as clay percentage increases (von Wandruszka, 2006). On the other hand, oxides retain P onto their surfaces in mono- or bi-dentate bonds. It is expected that in acidic soils, oxides ability to retain P increases, because oxide surfaces enhance their positive pH-dependent charge and also increase their reactivity (Yuji and Sparks, 2001). As a consequence, in calcaric soils both carbonate and oxides surfaces may have a role in P retention, while in acidic soils CaCO3 is nonexistent, and it is only the oxides that will mostly influence P. However, the role of oxides in P availability becomes complicated when soils at different stages of weathering are studied, because oxides content increases with weathering, but oxides crystallinity, a factor that reduces chemical reactivity, also increases (Shi et al., 2011). Moreover, it is expected that in soils with progressed weathering, parent materialderived CaCO3 should be diminished, and, if any exist, it should only be pedogenetic. It is thus unclear what soil properties will influence P retention in soils with similar pedogenetic “age” but with different pH, and what in soils differing in weathering. Although there have been various studies concerning the role of oxides to P retention, such works tend to study taxonomic orders of high degree of weathering (e.g., Oxisols, Ultisols), not usually found in, and thus not directly relevant

V. Antoniadis et al. / Catena 139 (2016) 214–219

to, the Mediterranean region. Also to our knowledge there is a void in the literature concerning works employing high number of soil samples (e.g., over 20), necessary for better statistical interpretation, that would compare soils divided according to pH (i.e., activity) and according to taxonomy and the degree of weathering. Thus it is important to study the role of oxides when comparing samples among (a) soils with variable degree of weathering, where the role of oxides may be more predominant, and (b) soils within the same taxonomy class, where the role of pH and CaCO3 may be more predominant. In such a manner, it would be possible to clarify the role of soil properties in P retention within each of these soil categories. This is even more evident in areas with low organic-matter content (such as those in the Mediterranean region). Organic matter is known to affect P sorption, as was also observed by Debicka et al. (in press), who found that removal of organic matter with H2O 2 decreased P sorption capacity and increased P desorption. In our tested soils, organic C is expected to have no significant effect. We tested the hypothesis that in soils at different stage of pedogenetic age, P sorption will be dependent on oxides, while in recently developed soils, CaCO3 will control P retention. Thus the aim of this work was to study P mobility in low organic matter soils, by using a series of soil extractants and batch sorption isotherms, in order to examine the role of various parameters in soils with variability in the degree of weathering and pH. 2. Materials and methods We obtained 23 soil samples at 0–20 cm depth from Central Greece, continuously cultivated for decades. Ten of them were of the same taxonomic order (Entisols) and selected so that they may differ in activity: Five of them (soils A, B, C, D, and E) were acidic, while the rest (soils F, G, H, I, and J) were alkaline, either Xerofluvents (Fluvisols according to World Reference Base (WRB) classification (FAO, 2015), soils of xeric temperature regime, developed on fluvial materials which is being deposited more frequently than soil development process rates, thus without a B horizon) or Xerorthents (Regosols according to WRB, xeric, erosion-affected shallow soils with no B horizon). The other 13 soils represented the three major taxonomic orders of different progress in weathering typical in the Mediterranean: 4 were Entisols (all of which Xerofluvents, soils E1, E2, E3, and E4), 5 Inceptisols (Cambisols as per WRB, xeric, with intermediate level of weathering, typically with ochric epipedon and cambic B horizon, soils I1, I2, I3, I4, and I5), and 4 Alfisols (Luvisols as per WRB, the most highly weathered soils of the region, all of which Rhodoxeralfs, otherwise known as “Red Mediterranean soils” or “Terra Rossa”, soils A1, A2, A3, and A4). In the 13 taxonomy-divided soils pH was not a selection factor. Details on the exact geographical positioning, as well as the taxonomic class of the obtained samples, are presented in the supplementary material (Table S1, and a “.KML”extension file in Google Earth background). All 23 soil samples were purposefully obtained from cultivated areas so that they may be of low organic matter. The samples were air-dried, passed through a 2mm sieve, and analysed for selected physico-chemical properties according to Rowell (1994): pH (1:2.5 H2O), particle size distribution (Bouyoucos hydrometer), organic C (wet oxidation), CaCO3 (calcimeter), and cation exchange capacity (CEC, 1 M CH3COONa). Soil oxides were also measured: total “free” Al and Fe oxides (pH-buffered dithionite-citrate-carbonate method, annotated with subscript “d” thereafter), as well as the amorphous oxides (extracted with ammonium oxalate, with subscript “o” thereafter). From the difference of the two, we estimated the well-crystalline, with subscript “d–o” thereafter. We also conducted four P extractions: Water soluble (10 mM CaCl2, thereafter WS-P), ammonium oxalate (AO-P), Mehlich-3 (M3-P), and Olsen (with 0.5 M NaHCO3, Olsen-P). We also calculated phosphorus saturation index (PSI) as the fraction of AO-P over the sum of Alo and Feo (all units in mmol kg−1). Then we conducted batch sorption tests at 1-to-10 soil-to-solution ratio with added phosphorus concentrations

215

of C0 = 1–100 mg L−1. We measured P sorption, q, and P in the equilibrium solution, C, and we fitted the experimental data to the Freundlich and Langmuir isotherms, as follows: q ¼ K F C N ðFreundlichÞ; linearized as logq ¼ logK F þ N logC; and q ¼ q max K L C=ð1 þ K L C Þ ðLangmuirÞ; linearized according to Lineweaver–Burk as 1=q ¼ ð1=q max K L Þ ð1=C Þ þ 1=q max ;

where KF N, qmax and KL are constants, with the latter two related to maximum sorption capacity and bonding strength, respectively. In order to assess the goodness of fitness, we used R2 (the closer to unity the better the fitness) and an error function (the lower the value, the better the fitness): Derivative of Marquardt's Percent Standard Deviation (MPSD, as per Foo and Hameed, 2010). From the fitted isotherms we calculated the distribution or partitioning coefficient (Kd), averaged over the whole range of added P concentration. This sorption index, along with q100 (measured sorption at C0 = 100 mg L−1) and qmax (derived from Langmuir) were used in this work for studying P. We then performed regression analyses between soil properties versus P extractability and P sorption indices. Also average values in various soil divisions were compared for their significance at the level of p b 0.05 with ANOVA. All such analyses were conducted with the use of the statistical package Statgraphics. We present the above mentioned parameters of soil properties, P extractability, and soil oxides content, and P sorption isotherm parameters, and statistically compare them as averaged for the various studied soil divisions, but we also include the data per soil Tables S2-S4 (supplementary material). 3. Results Soil properties did not differ among soil orders in the taxonomydivided soils, except for clay and CEC, which were significantly higher in Alfisols. In the activity-divided soils, pH and CaCO3 were significantly higher in the alkaline soils, as was also the case with organic C. Acidic soils also extracted higher Olsen-P and M3-P levels than the alkaline (Table 1). Although the percentage of amorphous-over-total free Al oxides was higher in Entisols, amorphous oxides were not different among soil orders (Table 2). As for total free dithionite-extracted oxides, they were higher in Alfisols. Phosphorus saturation index, indicating P retained onto chemically reactive oxides, did not differ in any of the three soil orders, a reflection of the similar levels of AO-P, Alo and Feo across soil orders, and neither did it differ in the activity-divided soils. Phosphorus sorption (Fig. 1, where for better comparison x- and yaxis dimensions were kept the same for all graphs), was rather higher in Alfisols (Fig. 1c) than in Entisols (Fig. 1b). This trend was also confirmed by the sorption parameters calculated with Freundlich and Langmuir (Table 3). Freundlich-calculated Kd was indeed higher in Alfisols, but qmax and q100 were not different. Both isotherm models were good in predicting P sorption (as indicated by R2 values approaching unity), but the MPSD error function revealed that Freundlich had lower error values. In the taxonomy-divided soils, Kd was significantly and positively correlated with clay (R2 = 0.71, p = 0.0003) and CEC (R2 = 0.83, p b 0.0000), and negatively (indicating a reversely proportional relationship) with sand (R2 = −0.46, p = 0.0104). q100 was negatively correlated with WS-P (R2 = −0.46, p = 0.0109), while qmax increased with Olsen-P and M3-P (Table 4). In the pH-divided soils, q100 and Kd increased with CaCO3 and with pH, while they decreased with enhanced M3-P levels. In the taxonomy-divided soils, both Al and Fe amorphous oxides influenced P sorption as indicated by Kd (R2 = 0.52, p = 0.0053 for Alo and R2 = 0.36, p = 0.0304 for Feo), while no such influence was observed in the pH-divided soils (Table 5). Total free dithioniteextracted Al and Fe oxides generated significant relationship with Kd,

216

V. Antoniadis et al. / Catena 139 (2016) 214–219

Table 1 Average values of the properties of the taxonomy-divided and of the activity-divided soils. p-values are reported in ×103 for clarity. a

¶

OC (%) Clay (%) Sand (%) CaCO3 (%) pH CEC (cmolc kg−1) WS-P¥ (mg kg−1) Olsen-P (mg kg−1) M3-P‡ (mg kg−1) AO-P† (mg kg−1)

Alfisols

Entisols

Inceptisols

p

0.79 37.3b 26.6 0.13 7.05 26.36b 0.370 54.83 63.43 325.91

0.94 28.9ab 48.4 1.44 7.39 18.81ab 0.517 58.20 119.48 321.95

0.84 22.4a 44.2 0.63 7.38 8.38a 0.619 59.24 130.21 341.59

757 10.0 209 285 531 79.5 967 909 452 111

NS¶¶ ⁎ NS NS NS ⁎ NS NS NS NS

b

¶

OC (%) Clay (%) Sand (%) CaCO3 (%) pH CEC (cmolc kg−1) WS-P¥ (mg kg−1) Olsen-P (mg kg−1) M3-P‡ (mg kg−1) AO-P† (mg kg−1)

Acidic

Alkaline

p

0.80a 39.5 39.8 0.00a 5.70a 9.75a 0.151 31.78b 68.65b 251.0

1.51b 39.0 35.8 7.70b 7.70b 19.75b 0.147 10.03a 43.77a 402.8

8.1 923 598 11.0 0.1 27.2 967 25.8 41.8 111

⁎⁎ NS NS ⁎ ⁎⁎⁎ ⁎ NS ⁎ ⁎ NS

Different letters within lines denote significant differences at the p b 0.05 level. ¶ Organic C. ¥ Water-soluble P. ‡ Mehlich-3-extractable P. † Ammonium oxalate-extractable P. ¶¶ Non significant. ⁎ Significant at p b 0.050. ⁎⁎ Significant at p b 0.010. ⁎⁎⁎ Significant at p b 0.001.

as also did even the well-crystalline Fed-o oxides (but no significance was found for Ald-o). 4. Discussion 4.1. Soil properties and oxides content CEC in the alkaline soils was significantly higher than that in the acidic (Table 1), and this was connected to pH, because even in soils Table 2 Average values of the oxides contents of the taxonomy-divided and of the activity-divided soils. Oxalate-extractable oxides are marked with subscript “o,” dithionite-extracted with “d,” and the ratio of oxalate-over-dithionite with “o/d.” p-values are reported ×103 for clarity. Alo

Feo

Ald

Fed

Alo/Ald

Feo/Fed

mmol kg−1

PSI¥ %

Taxonomy-divided soils Alfisols 34.55 Entisols 43.70 Inceptisols 28.06 p-Value 481 Significance NS¶

32.87 55.48 28.87 177 NS

90.06b 28.01a 32.72a 0.9 ⁎⁎⁎

180.51b 69.16a 61.07a 0.7 ⁎⁎⁎

0.39a 1.64c 0.93b 0.1 ⁎⁎⁎

0.18a 0.81b 0.50b 41.9 ⁎

32.87 55.48 28.87 177 NS

Activity-divided soils Acid 25.07b Alkaline 16.71a p-Value 10.9 Significance ⁎

23.67b 15.67a 39.6 ⁎

12.33 9.15 70.0 NS

24.02 16.13 117 NS

2.05 1.86 331 NS

1.02 0.98 661 NS

27.03 25.66 864 NS

Different letters within columns denote significant differences at the p b 0.05 level. ¥ Phosphorus saturation index. ¶ Non significant. ⁎ Significant at p b 0.050. ⁎⁎⁎ Significant at p b 0.001.

Fig. 1. Phosphorus sorption isotherms of (a) Alfisols, (b) Entisols, (c) Inceptisols, (d) acidic, and (e) alkaline soils. Graphs (a)–(c) include the taxonomy-divided soils, while graphs (d)–(e) include the activity-divided soils.

with predominantly permanent-charge clay minerals, such as those in newly developed activity-divided Entisols studied here, overall charge per kg soil (and thus CEC) decreases with pH due to variable pHdependent charge colloids (i.e., oxides and organic matter, Jiang et al., 2015). As for organic C, its lower levels in the acidic soils probably reflects the lower productivity of acidic soils, which results in lower biomass inputs to soil. Higher Olsen-P and M3-P levels in the acidic soils seem to suggest more generous P fertilizer application, so that farmers may compensate for lower productivity. WS-P (representing P

V. Antoniadis et al. / Catena 139 (2016) 214–219 Table 3 Average values of sorption parameters of the taxonomy-divided soils and of the activitydivided soils according to Langmuir and Freundlich. p-values are reported in ×103 for clarity. a Alfisols

Entisols

Inceptisols

p

Observed q100¶ (mg kg−1)

562

503

470

273

NS¶¶

Langmuir qmax† (mg kg−1) KL (L mg−1) R2 MPSDL‡

418 0.260 0.979 0.4286

488 0.265 0.969 0.4927

818 0.032 0.985 0.1584

221 276 432 117

NS NS NS NS

Freundlich N KF (mg kg−1) R2 MPSDF Kd¥ (L kg−1)

0.567a 69.38b 0.987ab 0.0697ab 37.36b

0.692ab 44.08ab 0.981a 0.1269b 31.84ab

0.827b 19.46a 0.995b 0.0348a 25.14a

32.6 20.5 54.4 58.7 44.1

⁎ ⁎ ⁎ ⁎ ⁎

b Acidic

Alkaline

p

Observed q100¶ (mg kg−1)

374a

541b

6.8

⁎⁎

Langmuir qmax† (mg kg−1) KL (L mg−1) R2 MPSD‡

1339 0.040a 0.923 0.3932

435 0.163b 0.946 0.2303

208 28.5 549 143

NS ⁎

Freundlich N KF (mg kg−1) R2 MPSD Kd¥ (L kg−1)

0.759b 18.75a 0.916a 0.2488b 8.33

0.543a 63.38b 0.967b 0.0697a 24.60

28.2 4.4 10.9 5.0 71.5

⁎ ⁎⁎ ⁎ ⁎⁎

NS NS

217

intensity, depending thus on P buffering capacity, rather than P quantity, as per Mejias et al., 2013) was not different among soil groups, as was also the case with AO-P, which tends to extract higher P levels, of similar magnitude to total phosphorus (Wuenscher et al., 2015). Well-defined crystal oxides structure develops with time, and thus well-crystalline oxides are usually found in soils at a more progressed weathering stage. This is agreed by Uzarowicz (2013), who investigated oxides crystallinity in a series of chronosequence soils spanning a nearly thousand-year period, but it also concurs with Nielsen et al. (2014), who reported data from a 4-year in situ soil ageing process. Indeed amorphous-over-total free oxides approached unity in Entisols, meaning that nearly all oxides were still amorphous, while in Alfisols amorphous Fe oxides were 18% of the total free oxides (significantly lower than Entisols, p b 0.001, Table 2). This confirmed the higher level of progressed weathering that has occurred in this soil order (as also agreed by Shaheen, 2009; Peng et al., 2015). In the 10 pH-divided soils, none of the measured oxides indices differed between the acidic and the alkaline soils, except for the oxalateextracted oxides, which were significantly higher in the acidic soils. This shows that chemically reactive amorphous oxides do not decrease only with progressing soil development, but also with soil alkalinity within same soil orders, because net positive surface charge greatly decreases with increasing pH, and so do free ionic Al and Fe species. This was also agreed by Ishiguro et al. (2006), who reported that anion (in that work, SO24 −) sorption was greatly reduced with increasing pH from 4 to 7 in an amorphous oxide-rich soil, due to a decrease in positive surface charge. 4.2. Phosphorus sorption

NS

Different letters within lines denote significant differences at the p b 0.05 level. ¶ Sorption at the maximum added metal concentration of C0 = 100 mg L−1. † Maximum sorption as calculated by the Langmuir model. ‡ Derivative of Marquardt's Percent Standard Deviation error function (MPSD). ¥ Partitioning coefficient averaged over the whole range of added P concentration of C0 = 0–100 mg L−1. ¶¶ Non significant. ⁎ Significant at p b 0.050. ⁎⁎ Significant at p b 0.010.

Phosphorus sorption (Kd, Table 3) was higher in Alfisols probably due to higher content of total oxides, known to affect P sorption (as also reported by Igwe et al., 2010). High clay content in Alfisols is also likely to have a role in P retention, as was also reported by Zhou and Li (2001), who found that “P sorption at relatively high solution concentrations […] appears to be caused by the affinities of […] noncarbonated clay.” Also P sorption was lower in acidic soils (Fig. 1d) than in alkaline (Fig. 1e), reflecting thus the role of CaCO3 in P retention. Contrary to what was expected, q100 and qmax were not higher in Alfisols. It is possible that qmax here does not directly predict sorption affinity, but it is rather affected by KL, a parameter representing binding strength, which was higher (but still not significant) in Alfisols and

Table 4 Coefficient of determination, R2, of the relationships of P sorption indices (q100, qmax, and Kd) versus soil properties and P extractions. Negative sign denotes adversely proportional relationship. All p-values (in parentheses, reported in ×10−3 for clarity) for an R2 N 0.20 are reported, but only the relationships with p b 0.05 are significant. OC¶

Clay

Sand

CaCO3

pH

% Taxonomy-divided soils 0.00 q100 (mg kg−1) qmax (mg kg−1) Kd (L kg

−1

)

qmax (mg kg Kd (L kg−1) ¶

−1

)

‡

0.01

0.83⁎⁎⁎

−0.28 (64.9)

−0.43⁎ (14.8) 0.71⁎⁎⁎ (0.3)

0.07

0.06

0.18

0.00

−0.14

0.00 0.25 (143)

0.07 0.01

0.00 −0.30 (101)

Organic carbon. Water-soluble P. Mehlich 3-extractable P. † Ammonium oxalate-extracted P. ⁎ Significant at p b 0.05. ⁎⁎ Significant at p b 0.01. ⁎⁎⁎ Significant at p b 0.001. ¥

0.14

0.01

0.10

mg kg

(10.4)

(0.0) 0.56⁎ (12.3) −0.03 0.48⁎

0.71⁎⁎ (2.4) −0.04 0.61⁎⁎

(26.7)

(7.4)

0.28⁎⁎⁎ (0.3) 0.02 0.18

−0.11 0.19 0.01

Olsen-P

M3-P‡

AO-P†

0.06

0.00

0.08

0.40⁎ (21.0) −0.11

0.71⁎⁎⁎ (0.3) −0.19

0.28 (61.0) −0.04

−0.35 (71.7) 0.04 −0.33 (83.0)

−0.61⁎⁎ (7.7) 0.20 −0.53⁎

−0.36 (64.9) 0.02 −0.40⁎ (48.6)

−1

−0.46⁎ (10.9) 0.00

0.00

−0.46⁎

−1

0.30 (52.4) −0.09

0.10

0.01

Activity-divided soils q100 (mg kg−1)

cmolc kg 0.17

0.09

WS-P¥

CEC

(16.5)

218

V. Antoniadis et al. / Catena 139 (2016) 214–219

Table 5 Coefficient of determination, R2, of the relationships of P sorption indices (q100, qmax, and Kd) versus oxides content in soils. Oxalate-extractable oxides are marked with subscript “o,” dithionite-extracted with “d,” and the difference between oxalate- and dithiote-extractable with “d–o.” Negative sign denotes adversely proportional relationship. All p-values (in parentheses, reported in ×10−3 for clarity) for an R2 N 0.20 are reported, but only the relationships with p b 0.05 are significant. Alo mmol kg

Feo

Mno

Ald

Fed

Ald-o

Fed-o

−1

PSI %

Taxonomy-divided soils q100 (mg kg−1)

0.08

0.15

0.11

0.19

qmax (mg kg−1)

−0.05

−0.12

−0.06

−0.08

Kd (L kg−1)

0.52⁎⁎ (5.3)

0.36⁎ (30.4)

0.26 (75.3)

0.49⁎⁎ (7.7)

Activity-divided soils q100 (mg kg−1)

−0.09

0.00

qmax (mg kg−1) Kd (L kg−1)

0.00 −0.08

−0.24 (147) 0.00 −0.19

−0.16 0.00

0.15

0.16

−0.05

−0.11

−0.40⁎ (20.4) 0.12

(41.3)

0.25 (63.2)

0.33⁎ (40.9)

−0.54⁎⁎ (4.0)

0.01

−0.11

0.16

0.02

−0.16

−0.02 0.00

0.00 −0.12

−0.12 0.27 (126)

−0.10 0.00

0.00 −0.27 (122)

0.28 (65.3) −0.20 (125) 0.59⁎

⁎ Significant at p b 0.05. ⁎⁎ Significant at p b 0.01.

Entisols than in Inceptisols. It is known that higher KL is related to slower retention processes, and that usually KL is inversely proportional to qmax, meaning that the slower the retention process, the higher the expected sorption maxima (as also agreed by Wang and Liang, 2014; Afsar et al., 2012). We thus assume that this is the reason qmax did not follow the trends of Kd. This also was the case in the activity-divided soils, where qmax was higher for lower KL, but still not different between acidic and alkaline soils. 4.3. Correlation analyses Regression (Table 4) showed that higher clay- and lower sandcontent soils (and thus higher CEC soils) had higher affinity for P sorption, even though P is an anionic species in soil solution (mostly found − as HPO2− 4 and H2PO4 , Molete et al., 2008). In the pH-divided soils no relationship was generated due to the fact that clay and sand contents happened to be in a very narrow range in those soils (Table 1). q100 was correlated with WS-P, indicating that the higher the initial P levels, the lower the expected P sorption soil capacity (Brennan and Bolland, 2003). qmax showed the opposite trend to that of other sorption indices (i.e., unlike Kd, it increased with decreasing clay content, and, unlike q100, it increased with Olsen- and M3-P). The discrepancy between qmax and the other sorption indices is probably caused by the same factor as that observed earlier, i.e., that qmax increases with increasing bonding strength (and thus with decreasing KL), rather than expressing P sorption affinity in soil. This was also confirmed in the activity-divided soils, where qmax, contrary to the other sorption indices, did not generate any significant relationship with any of the studied soil parameters. Sorption also significantly increased with pH and CaCO3 (as also found by Naeem et al., 2013). This shows that in soils with similar degree of weathering, CaCO3 plays the most important role in P retention, and thus sorption is expected to increase with soil alkalinity, while in soils differing in the level of development, other parameters are important. In the taxonomy-divided soils, amorphous, as well as well-crystalline oxides significantly influenced P sorption (Table 5). This concurs with Jan et al. (2013), who found that P was sorbed more onto amorphous Al and Fe oxides. They also reported that crystalline Fe sorbed up to 6 times less P than amorphous Fe, while crystalline Al oxides up to 44 times less P than amorphous Al, indicating thus that crystalline Al were a lot less capable of retaining P than crystalline Fe. This is the probable reason why we found a significant relationship of P with crystalline Fe, but not with crystalline Al. Similarly, PSI seemed to have influenced P sorption in the taxonomydivided soils, but not in the pH-divided soils. This showed the significance of the degree of the saturation of P onto amorphous oxides phases,

indicating that the lower the P saturation, and thus the PSI, the higher the P retention potential of a soil, as also agreed by Zhou and Gao (2011). 5. Conclusions Soil taxonomy is a key factor for understanding P mobility: Soils with more progressed development (even Alfisols, which, typical to soils in the Mediterranean region, are not highly weathered compared to other soils, e.g., in the tropics), are of higher ability to retain P compared to newly developed (here, Entisols) and of intermediate weathering (here, Inceptisols) soils. P sorption is mainly influenced by soil oxides, either amorphous, well-crystalline, or total free, in soils of different level of weathering (even in cases where these soils greatly differ in pH and CaCO3), while within the same weathering class, CaCO3 becomes the decisive factor that influences P retention. Acknowledgements We wish to thank Dr. Christos D. Tsadilas, Director of the Institute of Soil Mapping and Classification in Larisa, Greece, for providing us detailed (and accurate) soil maps which helped us considerably in obtaining correct soil samples. We also wish to thank Dr. Dionisios Gasparatos, Assistant Professor of Aristotle University of Thessaloniki, Greece, for helping us with his valuable in-depth knowledge concerning soil oxides, and for providing us the methods for extracting and analysing them. Map. KML file containing the Google map of the most important areas described in this article. Appendix A. Supplementary data Supplementary data associated with this article can be found in the online version, at http://dx.doi.org/10.1016/j.catena.2016.01.008. These data include the Google map of the most important areas described in this article. References Afsar, M.Z., Hoque, S., Osman, K.T., 2012. A comparison of the Langmuir, Freundlich and Temkin equations to describe phosphate sorption characteristics of some representative soils of Bangladesh. Int. J. Soil Sci. 7, 91–99. Al-Rohily, K.M., Ghoneim, A.M., Modaihsh, A.S., Mahjoub, M.O., 2013. Phosphorus availability in calcareous soil amended with chemical phosphorus fertilizer, cattle manure and sludge manure. Int. J. Soil Sci. 8, 17–24.

V. Antoniadis et al. / Catena 139 (2016) 214–219 Brennan, R.F., Bolland, M.D.A., 2003. Soil properties as predictors of yield response of clover (Trifolium subterraneum L.) to added P in soils of varying P sorption capacity. Aust. J. Soil Res. 41, 653–663. Debicka, M., Kocowicz, A., Weber, J., 2015. Organic matter effects on phosphorus sorption in sandy soils. Arch. Agron. Soil Sci. http://dx.doi.org/10.1080/03650340.2015. 1083981 (Article in press). FAO, 2015. World Reference Base for Soil Resources 2014: International Soil Classification System for Naming Soils and Creating Legends for Soil Maps Update 2015 www.fao. org/3/a-i3794e.pdf (accessed on 22th of December 2015). Foo, K.Y., Hameed, B.H., 2010. Insights into the modeling of adsorption isotherm systems. Chem. Eng. J. 156, 2–10. Igwe, C.A., Zarei, M., Stahr, K., 2010. Fe and Al oxides distribution in some Ultisols and Inceptisols of southeastern Nigeria in relation to soil total phosphorus. Environ. Earth Sci. 60, 1103–1111. Ishiguro, M., Makino, T., Hattori, Y., 2006. Sulfate adsorption and surface precipitation on a volcanic ash soil (allophanic Andisol). J. Colloid Interface Sci. 300, 504–510. Jan, J., Borovec, J., Kopacek, J., Hejzlar, J., 2013. What do results of common sequential fractionation and single-step extractions tell us about P binding with Fe and Al compounds in non-calcareous sediments? Water Res. 47, 547–557. Jiang, J., Yuan, M., Xu, R., Bish, D.L., 2015. Mobilization of phosphate in variable-charge soils amended with biochar derived from crop residues. Soil Tillage Res. 146 (Part B), 139–147. Mejias, J.H., Alfaro, M., Harsh, J., 2013. Approaching environmental phosphorus limits on a volcanic soil of Southern Chile. Geoderma 207-208, 49–57. Molete, S.F., du Preez, C.C., Marake, M.V., 2008. Relations between phosphorus retention parameters and some other properties of benchmark soils in Lesotho. S. Afr. J. Plant Soil 25, 99–104. Naeem, A., Akhtar, M., Ahmad, W., 2013. Optimizing available phosphorus in calcareous soils fertilized with diammonium phosphate and phosphoric acid using Freundlich adsorption isotherm. Sci. World J. 680257 (article number). Nesme, T., Colomb, B., Hinsinger, P., Watson, C.A., 2014. Soil phosphorus management in organic cropping systems: from current practices to avenues for a more efficient use of p resources. In: Bellon, S., Penvert, S. (Eds.), Organic Farming, Prototype for Sustainable Agricultures. Springer, Dordrecht, pp. 23–45.

219

Nielsen, S.S., Kjeldsen, P., Hansen, H.C.B., Jakobson, R., 2014. Transformation of natural ferrihydrite aged in situ in As, Cr and Cu contaminated soil studied by reduction kinetics. Appl. Geochem. 51, 293–302. Peng, X., Yan, X., Zhou, H., Zhang, Y.Z., Sun, H., 2015. Assessing the contributions of sesquioxides and soil organic matter to aggregation in an Ultisol under long-term fertilization. Soil Tillage Res. 146 (Part A), 89–98. Rowell, D.L., 1994. Soil Science: Methods and Applications. Prentice Hall, Harlow. Shaheen, S.M., 2009. Sorption and lability of cadmium and lead in different soils from Egypt and Greece. Geoderma 153, 61–68. Shi, Z., Krom, M.D., Bonneville, S., Baker, A.R., Bristow, C., Drake, N., Mann, G., Carslaw, K., McQuaid, J.B., Jickells, T., Benning, L.G., 2011. Influence of chemical weathering and aging of iron oxides on the potential iron solubility of Saharan dust during simulated atmospheric processing. Glob. Biogeochem. Cycles 25, GB2010. http://dx.doi.org/10. 1029/2010GB003837. Uzarowicz, T., 2013. Microscopic and microchemical study of iron sulfide weathering in a chronosequence of technogenic and natural soils. Geoderma 197-198, 137–150. von Wandruszka, R., 2006. Phosphorus retention in calcareous soils and the effect of organic matter on its mobility. Geochem. Trans. 7, 6. Wang, L., Liang, T., 2014. Effects of exogenous rare earth elements on phosphorus adsorption and desorption in different types of soils. Chemosphere 103, 148–155. Wuenscher, R., Unterfrauner, H., Peticzka, R., Zehetner, F., 2015. A comparison of 14 soil phosphorus extraction methods applied to 50 agricultural soils from Central Europe. Plant Soil Environ. 61, 86–96. Yuji, A., Sparks, D.L., 2001. ATR-FTIR spectroscopic investigation on phosphorus adsorption mechanism at the ferryhydrite-water interface. J. Colloid Interface Sci. 241, 317–326. Zhou, H., Gao, C., 2011. Assessing the risk of phosphorus loss and identifying critical source areas in the Chaohu lake watershed, China. Environ. Manag. 48, 1033–1043. Zhou, M., Li, Y., 2001. Phosphorus-sorption characteristics of calcareous soils and limestone from the southern Everglades and adjacent farmland. Soil Sci. Soc. Am. J. 65, 1404–1412.